The extracellular matrix (ECM) is known to change markedly during tumor progression and its histological presentation is used as a prognostic indicator by pathologists. It is clear that the ECM provides both structural, physical and biochemical cues to both normal and tumor cells, with major effects on cell proliferation, survival, differentiation and motility - it is also crucial for regulation of angiogenesis. However, we do not have a good understanding of the biochemical composition of the ECM in tissues because of its complexity, crosslinking and insolubility that make biochemical analyses difficult. This group has developed innovative methods for analyzing the in vivo matrix of normal tissues and tumors by combining newly developed bioinformatic analyses of the ECM and ECM-associated proteins encoded in mammalian genomes with state-of-the-art proteomics analyses of ECM-enriched samples from mouse models of cancer and from human patient material from tumor banks. These methods and improvements proposed in this application will be used to investigate the composition and changes in ECM that occur during tumor progression, invasion, angiogenesis and metastasis. The nature and origins (tumor cells or stromal cells) of the tumor ECM will be characterized and associated growth factors and cytokines will be identified. The functional roles of proteins showing interesting changes during tumor progression will be analyzed using loss- and gain-of-function manipulations. These will be implemented using viral vectors developed in prior research by this group that allow manipulation of the levels of gene expression in specific cells either up or down. In addition to shedding light on the mechanisms of function of ECM proteins, these studies may identify targets for intervention in tumor progression. By comparing the profiles of ECM and ECM-associated proteins in human patient samples (normal and tumor tissues) and correlating those results with clinical data on tumor outcomes and response to therapy, it is hoped that it will be possible to identify